Detection of Currency Notes, Medicine Names and Object detection for Visually impaired using Deep learning

Objective and Purpose

Artificial Intelligence is outgrowing in demand in every sector in recent years. Deep Learning is playing an important role in solving complex problems related to images. Despite being in a digital world where UPI transactions have increased tremendously, However, there is a section of the population who are visually impaired people who are deprived of using digital transaction features. We have analyzed this problem and come to a consensus that deep learning can overcome the problems faced by visually impaired people. The model is more focused on recognizing the Indian currency note and medicine names as well as object detection. The additional feature of this model includes the audio output of the results. In this model, we have used a dataset of ‘Indian Currency Note’ which widely covers the currency notes which are in circulation as per RBI. For example, it excludes one rupee, two rupee, five rupee and one thousand rupee notes. The data is trained and tested which is one of the preliminary activities in order to achieve the accuracy of the model. Also we have adding medicine name detection in these. This can be used on the medicines and get the name of that medicine followed by an audio.

Local machine setup for working application

Requirements

◙ Text-Editor: Pycharm(recomend)/VScode/Spyder (any modern editor) Or Google Colab/Juypter Notebook

◙ Python >= 3.9

◙ Git >= 2.39.0.2 (Not mandatory)

◙ Linux/Windows

◙ For Windows users install torch from here https://pytorch.org/ and configure it using this link https://www.geeksforgeeks.org/install-pytorch-on-windows/

◙ For Window, it is installed in command line during installation using requirements.txt

Steps

➊ Visit Git repository from here https://github.com/pranaybedekar/Detection-of-Currency-Notes-Medicine-Names-and-Object-for-Visually-Impaired-using-Deep-Learning.git

➋ You can download code in zip format from above link or Run git clone (https://github.com/pranaybedekar/Detection-of-Currency-Notes-Medicine-Names-and-Object-for-Visually-Impaired-using-Deep-Learning.git) within terminal/CMD

➌ Create a virtual env for this project using this link https://docs.python.org/3/tutorial/venv.html and activate virtual environment for this project using this link https://docs.python.org/3/library/venv.html

➍ Open terminal/CMD in current working directory

➎ Now, Install dependencies using pip install -r requirements.txt . Ensure successfull installation of all dependencies

➏ Run detect.py in terminal/cmd for model ouput.

➐ If you want an Android application of it, then run ObjectDetectorHelper.kt You will get this (App/object_detection/android/app/src/main/java/org/tensorflow/lite/examples/objectdetection/ObjectDetectorHelper.kt)

➑ Now, Build all .gradle run and install the .Apk Boom! you did everything perfect 🌟

➒ To use app on your mobile, It's a offline app you don't need data for it.